Hi,

It is difficult to clearly predict off target sites actual cleavage since it can be highly sequence and locus accessibility dependent. crispr.mit.edu predicts off target loci but there will be a lot to check since in a 3x10e9-bp genome close relatives to a 20 bp sequence are quite probable. In some cases, a single mismatch will lead to a drop in sgRNA mutagenic efficiency whereas in other cases 3 to 4 can be tolerated. (This is the best reason to sequence the DNA to target to design the guides and not rely on databases gDNA sequences to avoid loss of efficiency on target du to SNP mismatches).

But first, the concern about off-target changes a bit depending on which model you want to apply your CRISPR/Cas9 tool on. If on cells, off-target is really going to be an important issue, if on mouse you can get rid of off-target mutations with backcross although it is pretty long (and as long as a probable off target does not affect the targeted zygotes development).

There are nevertheless some rules of thumb:

- mismatches close to the PAM (nt 11-20 of 5'-3') reduce Cas9 DNA cleavage activity more than mismatches in nt1-10 of guide sequence that are much more tolerated). I’d say similarly to PCR primers annealing mismatches in the very 3’ end.

- double nicking make off target cleavage drop below the detection level of NGS and individual off target sites that might be altered because of DSB when the guide drives wt Cas9 are not supposed to because a single guide bringing a single nickase should likely be far less mutagenic and repaired with high fidelity by base excision repair instead of NHEJ.

see Zhang and Church labs papers, Cong 2013, Ran 2013, Mali 2013 for these 2 points.

By the way, according to some DNA repair specialists NHEJ repair of DSB is not as mutagenic as what is looked for in CRISPR/Cas9 mutagenesis (and is very specific when properly controlled as in VDJ recombination). People propose that the presence of a sterically big Cas9 (~160 kDa) poised on the DNA or staying around after a DSB may accentuate the mutagenic NHEJ repair in CRISPR/Cas9 approaches. Thus, the same can be held true for the nickase that could in some cases promote a mutagenic event in the course of the repair of a SSB. All this to say that it is good to keep in mind that below the detection level of NGS doesn’t mean zero.

- using transfection or nucleofection of recombinant Cas9 and in vitro synthesized sgRNA leads to much less Cas9 in cells than when they are transfected with PX-series plamsids. See IDT documentation about their Alt-R system. Off target risk is thought to be reduced with less Cas9 for less time.

Finally, the best way to avoid trouble with off target is to apply good experimental design and control rules that have not changed with CRISPR novelty compared to the siRNA shRNA era. Use several guides, let’s say at least 2 that have appreciably different sequences and thus different off target sites and if generating clones, keep and characterize several of them with each guide.

Good luck!

Daniel

To predict off-target sites for your gRNA, you may use this web-tool: https://crispr.cos.uni-heidelberg.de/index.html